Deep Learning Healthcare Applications - Research Paper reads

Paper #1 - System for Recommending Facial Skincare Products

Key Notes

• Multi-feature processing and classification of skin quality and acne status

Approach

• K-means cluster to search for acne
• Binary threshold using an adaptive method
• Identified the location of acne and provided recommendations to consumers
• A scale of oiliness was produced by labeling images of the weighted average

Steps

• The first feature used to determine whether the skin is oily
• Acne detection
• The brightness image is subtracted from the normalized grayscale image

Paper #2 - A Computer Vision Application for Assessing Facial Acne Severity from Selfie Images

Keynotes

• Acne is the 8th most common skin disorder in the world
• Transfer learning approach by extracting image features using a ResNet 152 pre-trained model, then adding and training a fully connected layer to learn the target severity level from labeled images.
• Mobile application for acne assessment
• Extracted skin patches from facial skins
• Haar feature-based cascade classifier
• Eye location, we inferred the regions of the forehead, cheeks and chin skin patches

Paper #3 - Deep Learning Methods for Selecting Appropriate Cosmetic Products for Various Skin Types: A Survey

Key Notes

• The cosmetic data from various websites @cosme and @Nykaa gathered for this model evaluation.
• The cosmetic product composition will be given based on skin types; dry, natural or oily.

Paper #4 Deep Learning Algorithms for Recognition of Facial Ageing Features

Keynotes

• Wrinkles, Dark spots, Under-eye circles
• Face Detection, wrinkle detection, scoring
• Facial zone - ensemble of regression trees, retrained for 50 fiducial points (dlib implementation) + contours detection
• Alignment - affine transformation
• Wrinkles area detection - cut areas by support points

More Reads

• Computer-Aided Clinical Skin Disease Diagnosis Using CNN and Object Detection Models
• Real-time Segmentation and Facial Skin Tones Grading
• Classification of skin pixels in images

Paper #4 - Time-Series Anomaly Detection Service at Microsoft

Key Notes

• Spectral Residual (SR) - Approach based on Fast Fourier Transform (FFT). Key Steps are
• (1) Fourier Transform to get the log amplitude spectrum; 
• (2) calculation of spectral residual; and 
• (3) Inverse Fourier Transform that transforms the sequence back to the spatial domain

Visual saliency detection domain. Applying CNN on the basis of SR output directly

CNN as our discriminative model architecture

More Reads

Overview of SR-CNN algorithm in Azure Anomaly Detector

Github

Happy Learning!!!

Weekdays vs Weekends

• When you give an estimate on weekdays you have to include all the pressure that is going in you - meetings, team extension, blocking issues. It comes with a rough approach but clarity is not there as focus on too many deliverables or meetings
• When you pick the idea on weekend, You have the laser focus, no distraction, no bing message in chat, no meeting reminders. Sometimes all we need is a peaceful environment to get the right solution
• Estimates are great but having a peaceful and distraction-free environment is more important as most of it is mental logic, thinking. 

Keep going!!!

Weekend Reads

• Anti-aging startups - This area is getting good investments - Link
• Hacking with Circuit python - Link
• Dropbox ML Read - How ML saves us $1.7M a year on document previews
• WFH Here to stay - Why employers will have no choice but to embrace remote working 
• Have not worked much on services, API. During discussions with my friend this topic, API naming conventions hit me hard link

The new investments for Anti-Aging, Electric cars, Recycling supply chain are key upcoming trends towards sustainable innovation.

Keep Thinking!!!

Some New Perspectives

Sometimes few conversations strike you deeply. Assume I may live another 34 years, The number of days left is 12,410 days. I have already passed 14K days already. When I look in terms of days it seems halfway already gone. This was while attending one of the sessions I felt this thought. Another thought on routine vs rituals link. What we do with a sense of belonging vs doing for the sake of it. When we are able to connect see the big picture we know the impact of work and our involvement. I was thinking on the same lines. Not every moment we live, some moments we may live without being connected to the moment. We may just pass the days without actually living the day, moment. 

A bit more on learning, refocus, and finding my real interests in the coming days. This life is too short to plan for long.

• First 10 years - 3650 days - Going to school, learning to cycle, hardly remember anything now
• Second 10 years - 3650 days - Curiosity of school, tuitions, 10th, 12th, and getting into college
• Third 10 years - 3650 days - Fresher to evolve yourself, learn in breadth or depth or domain, It keeps going, going. Spending late long nights at work.
• Current 10 years - 3650 days - Doing masters mid 30. Figuring out next-level learning, Teaching, Travel, Consulting. Still loads to learn

When we summarize it looks like not every moment or everything is captured. Some small happy moments don't strike this moment. What we leave when we are not around is a big question. Hopefully, we can inspire others to be more positive, kind, help each other. Often it's a constant struggle to feel oneself deeply and connect everything both emotionally and rationally. Stay kind, Stay curious, Keep Going. What you like will become a ritual and part of your thoughts. What you don't like will always be a routine. It again depends on how much we can convince and be clear on the end goal, the big picture. Often, everyone does not see the same perspective/vision / the journey.

Coming out of perspectives. Next few books/topics to learn

• Designing Data-Intensive Apps
• Agile Conversations
• Building Evolutionary Architectures
• SRCNN
• Drool Engine Architecture

Learn more, live a more ritual life and less of routine life :)

Happy Learning!!!

The Rise of SaaS Domain focused Startups

2021 would see lot more core domain-based solutions. Cloud has become mainstream, now we will have more ready to use products cutting across domains. Broader DBT adoption across sectors. This will become more of ready to deploy solutions and one-stop-shop which has all required use cases onboarded.

• Retail cloud - Retail integration of suppliers, manufacturing, last-mile delivery
• Health cloud - Patient management, Disease management, Pre and post operations care, and medication
• Fin cloud - Lending, onboarding, customer integration, micro finances
• Edu cloud - Teaching, Testing, AI-based curriculum, follow-up, Feedback
• Insurance cloud - Insurance, mortgage, payments, Renewals

Keep Thinking!!!

Which App to Pick Whatsapp, Telegram ?

Threema (No Email, No Contacts, No Phone number) - Privacy at Rs.270/-. Better than telegram, WhatsApp, Signal.

Simplified Linkedin - Link

Newcomer social network - Link

With every app trying to get access to contacts, location, SMS. It has become even more hazardous and addictive to be lost. More than the merits the impact, personalization, unrestricted adult content are more threatening to mental health. We don't know how we spend time but time is spent with more time-consuming apps with zero productivity.

Keep Thinking!!!

Live a better offline Life!!!

Tracking @ 2021 - Paper Reads - Object Tracking

Paper - Object Detection and Tracking Algorithms for Vehicle Counting: A Comparative Analysis

Key Notes

• Models used - CenterNet and Deep SORT, Detectron2 and Deep
• SORT, and YOLOv4 and Deep SORT
• YOLO and SORT algorithms
• Counting Techniques
• Counting by frame differencing
• Counting by detection
• Motion based counting
• Deep learning based counting

OBJECT DETECTORS

CenterNet

• CenterNet functions on the intuition that if a detected bounding box has a higher Intersection over Union (IoU) with the ground-truth box
• CenterNet is a singlestage detector 

Detectron2

Detectron2 supports implementation to multiple object detection algorithms using different backbone network architectures such as ResNET {50, 101, 152}, FPN, VGG16

YOLOv4

You Only Look Once (YOLO) is the state-of-the-art object detection algorithm. New techniques adopted in YOLOv4 are: (i) WeightedResidual-Connections, (ii) Cross-Stage-Partial-Connections, (iii) Cross mini-batch, (iv) Normalization (CmBN), (v) Selfadversial-training, (vi) Mish-activation

EfficientDet

Follows single-stage detectors pattern

SORT

Simple Online and Realtime Tracking (SORT) is an implementation of tracking-by-detection framework where the main objective is to detect objects each frame and associate them for online and real-time tracking application

Deep SORT

A combination of Kalman Filter and Hungarian algorithm is used for tracking

Paper - Tracking Objects as Points

Key Notes

• We track objects by tracking their centers. We learn a 2D offset between two adjacent frames and associate them based on center distance.
• A simple displacement prediction, akin to sparse optical flow, allows objects in different frames to be linked
• Joint detection and tracking
• Early approaches [2, 47] used Kalman filters to model object velocities

Paper - SMOT: Single-Shot Multi Object Tracking

Key Notes

• The first stage exercises a per-frame object detector to localize object bounding boxes in each frame
• The second stage, tracklet generation, merges detection results to create a set of tracklets, i.e., short tracks, based on short-term cues

More Read

FairMOT: On the Fairness of Detection and Re-Identification in Multiple Object Tracking

Object Detection and Tracking in 2020

Paper - ByteTrack: Multi-Object Tracking

Key Notes

• ReMOTS: Self-Supervised Refining Multi-Object Tracking and Segmentation
• TransMOT: Spatial-Temporal Graph Transformer for Multiple Object Tracking
• Multiple Object Tracking with Correlation Learning

In this paper, we identify that the similarity with tracklets provides a strong cue to distinguish the objects and background in low score detection boxes

Data association is the core of multi-object tracking, which first computes the similarity between tracklets and detection boxes and then matches them according to the similarity.

Location and motion similarity are accurate in the short-range matching.

Code

Codes

• People Counter Link1, Link2
• Object Tracking with Opencv and Python, Link2
• Vehicle counting system built with Python/OpenCV, Link2
• GUI Traffic Count using Yolo, Link2

Happy Learning!!!

Sign off 2020

2020 was a very different year
It was challenging both personally and professionally
Professionally it was a long stretch to learn/work on time-bound projects
Teaching was ok-ok
Personally not so well, overall.
A big thank you to everyone with whom I shared my 2020. Apologies for anyone whom I may have hurt. At the end of 2020 in terms of memories are some vague travel moments, work from home always.

Keep Going!!!

Mental Health & AI - Research paper reads

Paper #1 - Forecasting the onset and course of mental illness with Twitter data

Key Notes

• Dataset - 204 individuals (105 depressed, 99 healthy).
• Models - Classify depressed and healthy content, supervised learning algorithms, time series model
• Depression­ related terms such as diagnosis, antidepressants, psychotherapy, and hospitalization

Paper #2 - Predicting User Emotional Tone in Mental Disorder Online Communities

Paper #3 - Mental Health and Sensing

Features

1. Facial expression
2. Heart rate variability
3. Eye movement
4. Electrodermal activity
5. Mobility and location
6. Speech patterns
7. Technology use
8. Activity
9. Social interaction
10. Communication patterns
11. Social media

More Reads

Big Data Analytics and AI in Mental Healthcare

What are We Depressed about When We Talk about COVID19: Mental Health Analysis on Tweets Using Natural Language Processing

Temporal Mental Health Dynamics on Social Media

Personal Mental Health Navigator: Harnessing the Power of Data, Personal Models, and Health Cybernetics to Promote Psychological Well-being

Keep Thinking!!!

Five Things for better User Privacy in Social Media, Ads, Google, Youtube

• Apply policy of Individuals have the right to request the restriction or suppression of their personal data in all the countries
• Explicit consent requires a very clear and specific statement of consent with the purpose
• Investigate on Statistical purposes of Data usage at Individual Level
• What level of data is aggregated at the user level. Apply the policy of restriction, archival
• Callout every Ads based on identifiable data that was used (Name, IP address, Location data)
• Warn them of Obsessive behavior but do not alert it or use it for advertising
• Advise them if you spot depression symptoms but do not remember the historical data more than a few months
• Have a more social approach toward better sex education, awareness, restrict on content but provide more relevant content, not addiction

After months of stalling, Google finally revealed how much personal data they collect in Chrome and the Google app. No wonder they wanted to hide it.
⁰
Spying on users has nothing to do with building a great web browser or search engine. We would know (our app is both in one). pic.twitter.com/lJBbLTjMuu

— DuckDuckGo (@DuckDuckGo) March 15, 2021

Keep Thinking!!!